1. Field
This invention relates to the field of digital image processing, particularly the reduction of row noise in CMOS image sensors.
2. Related Art
Complementary Metal Oxide Semiconductor (CMOS) image sensors suffer from row noise that is usually due to the random noise that shifts the voltage level(s) in Analog-to-Digital Converter (ADC) and is a well know problem within the field of digital imaging. The shift in voltage level affects the ADC slope and ADC output by creating an offset consistent in the entire row of pixels. Row noise is seen as randomly distributed horizontal lines that appear to be relatively darker or lighter than the surrounding background. The noise becomes much more apparent to the eye due to the high correlation of pixels along the horizontal.
In U.S. Patent (U.S. Pat. No. 6,646,681 B1) Macy, this problem is attacked by a location specific row-by-row based offset correction mechanism. This method estimates the offset caused by random level shifts in the ADC converter by passing through the entire row and the several rows above and below to collect statistics as depicted in FIG. 1. After collecting statistics and estimating an appropriate offset, the mechanism attempts to correct for the row noise by adding the same offset to each pixel in the row. Although this method can be effective in certain scenarios, there are several drawbacks associated with reducing row noise on a local-statistics row-by-row basis.
Row noise reduction based on localized row-by-row correction is not effective in eliminating any vertical non-uniformity in sensor response. Such non-uniformities in sensor response can emerge when imaging evenly illuminated scenes, such as skies or photography studio backdrops and show up as shift in light and/or color variations from top to the bottom of the image. The local row noise will be corrected but it will be corrected to the underlying non-uniformity of the sensor response. Also, since the estimated offset is determined for an entire row at a time based on local active area statistics that may contain bad pixels, the possibility for visible image defects due to offset estimation error is increased. The estimation of the offset is susceptible to characteristics of other noise or defects present and the content of the scene being imaged. Any error will look noticeably worse because of the linear correlation of the corrected pixels and the human eyes tendency to impose linear patterns onto an image. For instance, if there happens to be a higher concentration of bad pixels within a certain number of rows, then the statistics will be skewed and the offset estimation will be erroneous. Such error, as affected by bad pixels, will be exaggerated because of the high linear correlation of the row noise correction offset. The method of the current invention provides for a much more robust and reliable estimation of the necessary offset.
In addition to its possibility of offset estimation error, the local active area row-by-row method of row-noise correction has several other significant drawbacks associated with a CMOS imager design, operation and image quality. Such localized row-noise reduction methods require that several rows of pixels be stored in memory during a statistical analysis step thus increasing the amount of physical memory required, the time spent processing the data and the cost to manufacture.
Since localized row-by-row methods passes all lines and pixels of a sensor as if they have uniform color response, it is inappropriate for use with CMOS sensors with Color Filter Arrays (CFA). CFAs, such as the so-called Bayer pattern, pass only one color at each pixel. Since most image sensors have a higher sensitivity to red light than to green or blue light, the response of a pixel with a red filter will be much higher than that of a pixel with a either a blue or green filter even if there is an equal level of red and blue/green signal. Due to this drawback, row-by-row noise correction can only be applied after an interpolation or demosaic step that can dramatically change the statistics required to estimate the offset.
Thus, there exists a need to improve the offset estimation step of row-by-row row-noise reduction methods to reduce the possibility of exaggerating row noise, the cost and complexity of sensor implantation and the visual appearance of vertical variations in sensor response.